FORECASTING CONSUMPTION AND SUBSTITUTION OF SAWNWOOD PRODUCTS IN THE BUILDING INDUSTRY IN DAR ES SALAAM CITY, TANZANIA JOSEPH EXAUD MGANA

Transcription

1 FORECASTING CONSUMPTION AND SUBSTITUTION OF SAWNWOOD PRODUCTS IN THE BUILDING INDUSTRY IN DAR ES SALAAM CITY, TANZANIA JOSEPH EXAUD MGANA A DISSERTATION SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN FORESTRY OF SOKOINE UNIVERSITY OF AGRICULTURE, MOROGORO, TANZANIA. 2013

2 ii ABSTRACT The future consumption and substitution of wood products in the building industry is not well analyzed and examined hence demand for wood in the industry remain uncertain. The study on forecasting consumption and substitution of sawnwood in building industry was carried out in Dar es Salaam city. Consumption forecasts of sawnwood for years 2016, 2021 and 2026 were determined through income elasticity of demand forecasting model. The sampling unit was obtained from a list of wards, building contractors and architects in Dar es Salaam city. Random sampling with replacement was employed with an intensity of 20% - 100% depending on the population, availability of respondents and willingness to respond. Questionnaires were used to obtain data from interviewees. Statistical data of sawnwood and substitute building materials were recorded in data sheets. Data collected were analyzed using SPSS and MS Excel programmes. The consumption of sawnwood and substitute building materials were assessed in doors and windows. Buildings were categorized into Lower, Medium, and High categories. The lower category involved all none storey buildings and consumed an average of 2.69 m 3 per building unit while medium category (1-3 storey) consumed 3.1 m 3 of sawnwood per building unit. The last category involved buildings with 4 storeys and above which according to the results consumed an average of 5.3 m 3 of sawnwood per building unit. The study show that in year 2011, Dar es Salaam consumed a total of 8,706.9 m 3 of sawnwood for doors and window frames in about 2878 building units that were built in that same year. Kinondoni district consumed 42.2% of the total sawnwood while Ilala district consumed 34.8% of total sawnwood and Temeke district consumed 23%

3 iii of the total sawnwood consumed in windows and doors in Dar es Salaam city in the year The per capita building consumption of sawnwood for Dar es Salaam in 2011 was estimated to be 2.7 m 3 while for aluminium was 46.2 m 2. Sawnwood substitution was highly observed in windows compared to doors with aluminium being the main substitute material. The per capita building consumption of sawnwood in 2026 is estimated to be 3.4 m 3 which is an increase of 23.4% compared to 2.7 m 3 observed in year For aluminium, per capita consumption for 2026 will be about 86.8 m 2 which is an increase of 88% compared to 46.2 m 2 which was observed in year Sawnwood consumption in none storey buildings will still grow fairly fast because majority of the buildings in urban centers particularly Dar es Salaam are being built by low and middle income people for residential purposes. Therefore, the commercially unknown and underutilized sawnwood species need to be publicized on their strength properties, resistance to weather and durability for future consumption in the building industry.

4 iv DECLARATION I, Joseph Exaud. Mgana, do hereby declare to the Senate of Sokoine University of Agriculture that this dissertation is my own original work done within the period of registration and that it has neither been submitted nor being concurrently been submitted for a degree award to any other institution. Joseph Exaud Mgana Date (MSc Candidate) The above declaration is confirmed Professor Yonika. M. Ngaga Date (Supervisor)

5 v COPYRIGHT No part of this dissertation may be reproduced, stored in any retrieval system, or transmitted in any form or by any means without prior written permission of the Author or Sokoine University of Agriculture in that behalf.

6 vi ACKNOWLEDGEMENTS First and foremost I would like to thank the Lord Almighty God for my wellbeing and strength during the whole period of my study at Sokoine University of Agriculture. I would like to convey my deepest gratitude to the Belgium Technical Corporation (BTC) for the financial support without which my two years study at SUA could have not been possible. I am deeply grateful to my supervisor Prof. Y.M. Ngaga from the Department of Forest Economics for his guidance, positive criticisms and comments throughout the preparation and write-up of this dissertation. His diligent efforts in giving challenging advice have made the completion of this study possible. Regardless of having tight schedule, he always had time for my work. May God grant him with all the favour. Special thanks are also extended to Municipal Engineers of Dar es Salaam Region for releasing their information and direction on how to collect information from Building contractors and construction sites. I also thank Mr. David Solomon and all the wards executives for their assistance in data collection even in difficult raining conditions. I am also indebted by the Construction Registration Board (CRB), Engineers Registration Board (ERB), Architects and Quantity Surveyors Registration Board (AQSRB) for providing me with registered experts and their contacts which enabled

7 vii me to reach them without any resistance. Lastly, I would like to thank Mr. Grayson Nyamoga for devoting his time and resources to the fulfillment of this dissertation.

8 viii TABLE OF CONTENTS ABSTRACT... ii DECLARATION... iv COPYRIGHT... v ACKNOWLEDGEMENTS... vi TABLE OF CONTENTS... viii LIST OF TABLES... xiii LIST OF FIGURES... xv LIST OF PLATES... xvi LIST OF APPENDICES... xvii LIST OFABBREVIATIONS AND SYMBOLS... xviii CHAPTER ONE INTRODUCTION Background Information Problem Statement and Justification Objectives General Objective Specific Objectives Research Questions... 5 CHAPTER TWO LITERATURE REVIEW... 6

9 ix 2.1 Forest industry capacity and employment Building and construction industry Factors contributing to the increase in wood consumption Economic growth External trade Growth in construction industry Population growth Consumption forecasting Forecasting techniques Time series methods Single moving average Single exponential smoothing Linear moving average Brown s one parameter linear exponential smoothing Holt s two-parameter linear exponential smoothing Winter s linear and seasonal exponential smoothing Auto regressive model (AR) Moving average model (MA) Mixed autoregressive moving average model (ARMA) Causal forecasting methods Simple regression methods Multiple regressions Econometric models CHAPTER THREE... 21

10 x 3.0 METHODOLOGY Description of the study area Location Area, climate and landforms Area Climate Population Socio- economic activities Data collection Sampling procedure Primary data collection Questionnaires Direct observation Checklists and Key Informants Secondary data collection Data analysis Wood products consumption forecast Smoothing (average) technique Income elasticity of demand model CHAPTER FOUR RESULTS AND DISCUSIONS Consumption of building materials for doors and windows in Responses on sawnwood substitution in the building industry... 41

11 xi 4.3 Sawnwood products use in building Industry Sawnwood availability Sawnwood preference and demand in building industry in Dar es Salaam city Sawn wood prices in Dar es Salaam city Doors and windows prices Sawnwood doors and windows Aluminium doors and windows PVC windows and doors Factors underlying sawnwood substitution in the building industry Future sawnwood consumption in Dar es Salaam City Population Economic attributes Urbanization rate Forecast models chosen Time series Models Income elasticity of demand model Forecasted sawnwood consumption Forecasting substitution Total sawnwood consumption in the future CHAPTER FIVE CONCLUSION AND RECOMMENDATIONS Conclusion Recommendations... 80

12 xii REFERENCES APPENDICES... 90

13 xiii LIST OF TABLES Table 1: Natural forest area and harvesting potential in selected regions... 7 Table 2: Employment statistics in construction and other sectors in Tanzania for year Table 3: Consumption of sawnwood and substitute building materials in different building categories Table 4: Per capita consumption of sawnwood in building units from Table 5: Consumption of Sawnwood and Aluminium per district from year Table 6: The trend of consumption of sawnwood and aluminium in doors and windows in different building categories Table 7: Estimates by respondents on percentage range on aluminium application and preferences Table 8: Estimates by respondents on sawnwood substitution range in the building industry Table 9: Response on percentage estimates for sawnwood substitution in different building categories Table 10: Response on sawnwood use preferences Table 11: Sawnwood availability in Dar es Salaam Table 12: Responses on sawn wood demand in building and construction industry in Dar es Salaam city Table 13: Retail prices for hardwood sawnwood in Dar es Salaam city... 55

14 xiv Table 14: Stability of sawnwood prices in Dar es Salaam city Table 15: Market price estimates for Pterocarpus angolensis and Afzelia quanzensis sawnwood doors and window frames Table 16: Sawnwood consumption forecast using time series methods Table 17: Sawnwood consumption forecast in building and construction industry for Dar es Salaam from Table 18: A summary of sawnwood consumption forecasts using different models Table 19: Forecast of Aluminum Consumption based on IED Table 20: Total sawnwood consumption using different forecasting models... 78

15 xv LIST OF FIGURES Figure 1: Response awareness on presence of sawnwood substitution in the building industry Figure 2: Building materials substituting sawnwood in doors and window frames Figure 3: Responses on preferences for aluminium substitution Figure 4: Responses on aluminium availability in Dar es Salaam city Figure 5: Responses on sawnwood availability in Dar es Salaam city Figure 6: Sawnwood preferences in building industry in Dar es Salaam city Figure 7: Differences between real and current market prices of sawn hardwood species in Dar es Salaam Market Figure 8: Sawnwood doors and windows price changes from

16 xvi LIST OF PLATES Plate 1: Mar-kim Uni Plast Co. Ltd and workshop located in Dar es Salaam city Plate 2: Different designs of plastic window frames at Mar-kim Uni Plast Co. Ltd... 64

17 xvii LIST OF APPENDICES Appendix I: Questionnaire Appendix II: List of respondents... 96

18 xviii LIST OF ABBREVIATIONS AND SYMBOLS AIDS AR ARMA AQRB Acquired Immune Deficiency Syndrome Auto Regressive Auto Regressive Moving Average Architects and Quantity Surveyors Registration Board. BOT CRB DPG DRC DSM EPZ ERB FAO FBD FTE GDP HIV IED IMC IMR IPPC KMC Bank of Tanzania Construction Registration Board Development Partners Group Democratic Republic of Congo Dar es Salaam Export Processing Zone Engineers registration Board Food and Agriculture Organization Forestry and Beekeeping Division Full Technical Equivalent Gross Domestic Products Human Immunodeficiency Virus Income Elasticity of Demand Ilala Municipal Council Infant Mortality Rate International Plant Protection Conversion Kinondoni Municipal Council

19 xix m² Square Metre m³ Cubic Metre MA MDF MNRT MR MSE NBS NHC NSSF OSB PSPF PVC SPSS TAS TBA TIC TMC UNECE UNHABITAT URT USD VAT Moving Average Medium Density Fiberboard Ministry of Natural Resources and Tourism Mortality Rate Mean Square Error National Bureau of Statistics National Housing Corporation National Social Security Fund Oriented Strand Board Public Service Pension Fund Poly Vinyl Chloride Statistical Package for Social Science Tanzania Shilling Tanzania Building Agency Tanzania Investment Centre Temeke Municipal Council United Nations Economic Commission for Europe United nations Human Settlements Programme United Republic of Tanzania United States Dollar Value Added Tax

20 1 CHAPTER ONE 1.0 INTRODUCTION 1.1 Background Information Wood is one of the world's main building materials which is widely used in housing and construction activities. It can be sawn longitudinally, with or without its natural rounded surface with or without bark to produce sawnwood (IPPC, 2010). The use of wood-based panels and other wood products in construction and building works have shown an increasing trend hence the link between forest industries and building sector becomes evident (Unasylva, 1971). In Tanzania, the construction industry contributes more than 10% of GDP while its contribution to employment is reported to be more than 9% of the population (Ujenzisolutions, 2005). Mwampamba (2007) reported that the contribution of the construction industry in Tanzania for three consecutive years 2006, 2007 and 2008 were estimated to be , and TAS respectively. Other studies show that the construction sector is one of the fastest growing sector in the economy with an annual turnover of between 1.8 billion billion USD (Dailynews, 2011). Statistics show that construction projects with a total value of 2.8 trillion TAS were registered in 2010, with building works accounting for half of the total value of the projects. The value of the registered building works amounted to 1.4 trillion TAS, while civil works was 884 billion, specialists in electrical 231 billion TAS and electrical works 187 billion TAS (CRB, 2011). It is also estimated that more than 60% of the government budget is spent in the construction sector (ibid). The growth rate of the construction sector increased between 12-15% in 2009/10

21 2 compared to that of about 9% in 2008/09 (Tanzaniainvest, 2009). These changes in growth rate were attributed by an increase in the construction of residential and none residential buildings, roads, bridges and land improvement activities (ibid). The construction cost per square metre for houses varies dramatically based on site conditions, local regulations, economies of scale and the availability of skilled trade personnel. The rapid expansion of towns as a result of high rate of urbanization and commercial activities indicate an imbalance between the amount of wood products that can be supplied to the consumers and the actual requirements. Similarly, the emerging competition between wood products in the building and construction works with substitute materials like concrete, steel, plastic and aluminium may result into dwindling of wood markets (FAO, 2007). High quality reconstituted wood based panels such as particle board, medium density fiberboard (MDF) and oriented strand board (OSB) are predicted to reduce the consumption of locally made wood materials from traditional forest industries due to differences in tastes, preferences and quality (FAO, 2007). Forecasting the consumption of wood products under different driving forces is inevitable because many decisions for future development of the forestry sector will depend on the forces that influence the demand and supply of these wood products. 1.2 Problem Statement and Justification The quality, tastes, and prices of different building materials have resulted into a strong competition in the building sector. Preferences are changing among building

22 3 owners and they are switching from one species to the other. Some owners and building companies shift into none wood building materials. The substitution of wood with none wood materials or with different species causes a shifting demand of these building materials. The future consumption and substitution of wood products in building industry is not well analyzed and examined hence demand for wood in the industry remain uncertain. According to FBD ( 2011), there are replacements of wood framed ground floor system by concrete stab foundation, applications of roof trusses replacing sawnwood, plywood substituting sawnwood in roof sheathing and sub flooring, premature poles used in the formwork being replaced by metal poles and the use of aluminium to replace timber in door and window frames. Also, consumers have shifted into diverse species comprising of soft wood and lesser known hardwood species which were previously underutilized and ignored (Machumu, 2008; Zziwa et al., 2006b; Ishengoma et al., 2004). The consequences of these substitutions in the building sector are the decrease in the use of wood products in the sector and also reduced efforts of planting trees and weakening the local wood industries. The increase in demand for sawnwood in building industry will depend on the efficiency of the wood working industry and its ability to face competition from substitute materials and in this respect research and promotion will have to play an important role (Unasylva, 1971). Forecasting consumption and substitution in the building sector can be an efficient way of discovering the future consumption behaviour of wood products, the level of substitution and its consequences in the sector and to the forest industry at large.

23 4 Despite of the importance of the building sector in Tanzania, little information is available on future demand for wood products and its associated consequences as a result of substitution by other materials. The study on the forecasting consumption and substitution of wood products in this sector provides an insight on wood products market situations and the general trends in the market. This information is useful to stakeholders such as tree growers, wood traders and other stakeholders closely related to the forest sector. Findings from this study will serve as a basis for promoting use of wood products by designers and builders. Moreover, findings on substitute building materials may stimulate trades and promote local industries as well as contributing to environmental and forest conservation. The study also provides some insights on the future consumption of sawnwood products and the rate of substitution in the building industry in Dar es Salaam. 1.3 Objectives General Objective The general objective of this study was to investigate substitution of wood products and forecast its consumption in the building industry in Dar es Salaam city, Tanzania Specific Objectives The specific objectives were to: i. Estimate the present consumption of sawnwood products by the building industry. ii. Identify and assess the types of sawnwood products and areas being

24 5 substituted, and the level of substitution in the building industry. iii. Identify factors underlying substitution of sawnwood products by other materials in the building industry. iv. Forecast future consumption of wood products by the building industry. 1.4 Research Questions To answer the above specific objectives, the study was guided by the following research questions: i. What is the present consumption of sawnwood products in the building industry in Dar es Salaam? ii. iii. iv. What sawnwood products are substituted in the building industry and why? What is the level of substitution and factors underlying these substitutions? What is the future consumption of wood products by the building industry? v. What are the potential threats or benefits resulting from these substitutions?

25 6 CHAPTER TWO 2.0 LITERATURE REVIEW 2.1 Forest industry capacity and employment Tanzania has a total production forest area of about 23.8 million ha (FBD, 2011). The annual harvestable volume per annum in the existing production forests is estimated to be about 87.7 million m 3 while the annual volume harvested from natural forests in each district is between m 3 giving an estimate of m 3 of hardwood logs produced annually (ibid). The forest inventory report from 13 regions of Tanzania (Table 1) show that there is enough harvestable volume in the forest though it is largely from lesser known and commercially unknown species (FBD, 2005). The forest based industry in Tanzania is largely dominated by sawmills, furniture manufactures and other value added wood product manufactures (FBD, 2011). According to MNRT (2005), the country had about 367 sawmills in year 2005 with the installation capacity of m³ for softwood logs and m³ of hardwood logs per annum. However, its utilization capacity has not even reached half of the installation capacity. Most of these sawmills are reported to be small scale with input not exceeding 5000 m³ of logs and providing employment for only 5-8 skilled and unskilled employees per sawmill (ibid).

26 7 Table 1: Natural forest area and harvesting potential in selected regions Region Total Area, ha Harvestable area, ha Harvestable net volume, m 3 Morogoro Mbeya Rukwa Tanga Lindi Kagera Ruvuma Tabora Dodoma Coast Kigoma Iringa Shinyanga Total Source: FBD (2005) Forest sector contribute significantly in the provision of employment in developing countries including Tanzania. FAO (2011) reported that in 2006 Tanzania s forest sector provided an employment of about 15,000 skilled personnel (full technical equivalent). Round wood production employed about 3000 people while wood processing and pulp and paper each employed about 6000 people. On the other hand, DPG (2007) reported that about 1 million rural people are being employed by the forest sector. They further revealed that with proper data collection the statistics may exceed the reported ones. According to FBD (2000), the sector accounted for about people including men and women. Milledge et al. (2007) found that in

27 8 southern Tanzania, 16% of households living adjacent to natural forests benefited from timber trade and logging in Building and construction industry The building and construction sector in Tanzania is one of the most important sectors that boost the country's economy by transforming various resources into constructed physical, economic and social infrastructure. The construction and other sectors contribute to the creation of economic opportunities such as the ease of market access, increased competition, trade development, tourism and foreign investment, contribution to government revenues and employment opportunities (Tanzaniainvest, 2009). In year 2005, Dar es Salaam region accounted for 43.7% of all the employments provided in construction sector for urban areas while in rural areas the sector provides about employments of which are males and 1006 females as indicated in Table 2. Table 2: Employment statistics in construction and other sectors in Tanzania for year 2005 Sector Dar es Salaam Other Urban areas Rural areas Male Female Total Male Female Total Male Female Total Construction industry Other industries Total Source: Ujenzisolutions (2005)

28 9 Wood products substitution in the building and construction industry Solid wood products (sawnwood and wood based panels) are the potential building materials in most of the countries worldwide (FAO, 2011). The selection of construction materials is determined by among other factors energy, costs, durability and ease of use but these tend to decrease at high levels of income. Study by FAO (2011) show that substitution of sawnwood has also been driven by the decline in the quality and scarcity of good logs which can produce good quality sawnwood products. Milledge et al. (2007) found that in Dar es Salaam, the high urban demand for timber resulted into a depletion of most hard wood and more than 80% of the trees harvested within 20 km of the city were found to be used for house construction or charcoal production. 2.3 Factors contributing to the increase in wood consumption Most people would like to increase the quantity or quality of the goods they consume. They consume less than they desire because their spending is limited by their income. In the theory of consumer choice, Mankiw (2001) explained that the rate at which a consumer is willing to buy one good for the other depends on the satisfaction that a consumer receives from the goods which he or she is already consuming. On the other hand, Mankiw (2001) and Levin et al. (2004) describes that when the price of goods falls, consumer s choices can be impacted by income effect which is the change in consumption that arises because a lower price makes a consumer better off. Also, consumer s choices can be impacted by substitution effect described as the change in consumption which arises because a price change

29 10 encourages greater consumption of the goods that has become relatively cheaper. These expiations have been given with assumption that all consumers seek to maximize satisfaction from the combination of goods they wish to consume Economic growth There is a close relationship between growth rate of real Gross Domestic Product (GDP) and the growth rate of consumption of wood based products in the world. According to FAO (2007), countries with high real GDP consume more wood based products compared to those with low growth rate. Asia with a GDP growth rate of 15.8% in consumed 16.4 % of wood based panels compared to Asia Pacific with real GDP of 8% which consumed 10.2% of global consumption. On the other hand, Africa with a GDP growth rate of 2.2% consumed 1.0% of global consumption. Despite the close relationship between real GDP growth rate and wood based panel consumption, in most cases GDP does not take into account some of the economic activities hence underestimating the growth of economy of a particular country or region. The construction industry has continued to be one of the fast growing sector contributing significantly in the economic growth despite of the economic hardships experienced in many counties. The Tanzania building industry is currently experiencing a period of growth primarily driven by the recent developments in roads works, housing and mining industries. According to construction registration board (CRB), the construction sector has made an excellent performance in 2011 growing at around 12% of the GDP (Dailynews, 2011).

30 External trade Most countries with timber deficiency tend to import wood products to satisfy their local market demand. Countries which are rich in timber resources will expand its market by increasing exports. Study by FBD (2011) show that export of sawnwood in Tanzania increased from 511m³ in 2001 to m³ in 2007 while export of poles increased from 905 in 2004 to in However, sawnwood exports from Tanzania are facing challenges in the international market due to poor quality hence mostly are locally or domestically consumed Growth in construction industry Construction of residential and none residential houses remains to be the main activity which requires significant amount of wood based panels, either for concrete frame works, or for flooring, paneling and sheathing (FAO, 2007). FBD (2011) estimated that in Tanzania there are units built annually for both large scale industrial and commercial building activities. Dar es Salaam continued to register the largest number (76%) of valued construction projects in 2009 compared to other regions. This is an increase of 4.5% when compared to 71% of 2008 (CRB, 2011). Studies show that building and construction industry will continue to grow in urban centres which will cause an increase in demand for construction materials because of the increased number of investments in factories, manufacturing, processing industries and population (Shayo, 2006; FBD, 2011; UN, 2011). It will also increase due to increased newly built office premises, packaging materials and storage facilities.

31 Population growth The urban population growth rate in Tanzania is estimated to be around 4.2% which is two times compared to population growth rate in rural areas (FBD, 2011). The rural urban migrations results into an increased wood consumption for energy, building and construction materials. It is reported that Dar es Salaam has the second highest population growth rate of about 4.3% after Kigoma which has a growth rate of about 4.8%. This is also higher than country s average population growth rate. On the other hand 93.9% of the total region s population in Dar es Salaam lives in urban area (Unhabitat, 2009). The rapid increase in the population leads into growth of building activities and hence demand for building materials. 2.4 Consumption forecasting Forecasting is defined as quantitative estimates of some specified future conditions or events made as a result of rational study and analysis of available pertinent data (Gregory et al., 1971). There are some elements of uncertainties when economic decisions are made with respect to future events or conditions. Therefore the aim of forecasting is to try to reduce these uncertainties concerning the future conditions or events Forecasting techniques Several techniques which can be used for forecasting future conditions and events exist. In this study different techniques such as time series methods, causal forecasting methods and econometric models have been reviewed.

32 Time series methods The past history of a subject matter is an important part in application of these methods where by knowing it provides a room for extrapolation of its future behavior. Only variables which need to be forecasted are being explored for its past behavior to suit the models. Smoothing models base their forecasts on the principle of averaging (smoothing) past errors by adding a percent of the error to the percent of the former forecast. These methods differ depending on the way they are being used and the behaviour of data to be forecasted Single moving average Single moving average forecast for period t is given by: F t+1 = X t /N X t-n+1 /N + F t, (1) Where: F t+1 = The forecast for time t+1. X t = The most recent observation. N = The number of values included in the average. F t = Previous moving average Single exponential smoothing Single exponential smoothing forecast for period t is given by: F t + 1 = F t + a (X t F t ) (2) Where: a = the smoothing coefficient (lies between 0 and 1) F t+1 = the forecast for time t+1.

33 14 X t = the most recent observation. F t = Previous moving average Linear moving average The basis of this method is to calculate a second (double) moving average. Generally, linear moving average can be calculated using the following formulas: S t = X t + X t X t -N+1/N (3) S t = S t + S t -1 + S t -2 + S t N+1,... (4) A t = S t + (S t - S t ) = 2S t S t, (5) b t = 2 / N-1 (S t S t )... (6) The forecast is given by F t +M = A t + b t M (7) Where: S t = Single moving average. S t = Double moving average A t = Base adjustment to a starting point for a forecast. b t = Trend in the data at time t. M = Number of periods ahead to the forecast Brown s one parameter linear exponential smoothing This is one of the linear exponential smoothing methods. The underlying rationale of Brown s linear exponential is similar to that of linear moving averages using single and double smoothed values. Forecast can be made using the following formulas: S t`= αx t + (1-α) S t 1.. (8) S t = αs t + (1-α) S t (9)

34 15 A t = 2S t S t...(10) b t = α/1- α(s t S t )....(11) The forecast is given by F t + N = A t + b t M.. (12) Where: S t = Single exponential smoothed value. S t = Double exponential smoothed value. α = Smoothing coefficient for current smoothed level series N = The number of values included in the average Holt s two-parameter linear exponential smoothing This is another method under linear exponential smoothing. The forecast for Holt s linear exponential smoothing is found using two smoothing constants (with values between 0 and 1) and the following equations: S t = αx t + (1-α) (S t 1 + b t -1). (14) b t = μ(s t S t -1) + (1- μ)b t -1.. (15) The forecast is given by F t +N = S t + b t M (17) Where: S t = Exponential smoothing value at time t. S t -1 = Last exponential smoothing value. μ = Smoothing coefficient analogous to μ, b t = Smoothing trend in the data. b t 1 = Smoothed trend of the previous period.

35 Winter s linear and seasonal exponential smoothing Winter s linear and Seasonal exponential smoothing is one of the forecasting methods in time series. The method is capable of dealing with data series that contain a trend as well as seasonal pattern. This method involves the use of single smoothing parameter, trend smoothing parameter and seasonality smoothing parameter. The Basic equations for Winter s Linear and Seasonal Exponential Smoothing are: S t = αx t / I t-l + (1-α) (S t-1 +T t-1 ).. (18) T t = β (S t - S t-1 ) + (1-β) T t-1... (19) I t = γx t / S t + (1-γ) I t-l... (20) Where S t = Smoothed value of the deseasonalized series at time t T t = Smoothed value of the trend at time t I t = Smoothed value of the seasonal factor at time t L = The length of seasonality β = Smoothing coefficient for trend α = Smoothing coefficient for current smoothed level series γ = Smoothing coefficient for seasonality All smoothing coefficients lies between 0 and 1, and are obtained by trial and error to find the set of values which gives the minimum mean square error (MSE). X t / I t-l Eliminates seasonal fluctuations from X t.. To calculate the value of I t it first requires to know the value of S t. The incremental trend S t - S t-1 is weighted with β and the previous trend value T t-1 is weighted with ((1-β) as they appear in equation 2 above.

36 17 The forecast relied on winters method is computed by the following formula; F t+m = (S t + mt t ) I t-l+m... (21) Where m is the number of periods in which the forecast takes place Auto regressive model (AR) In this model the current observation depends on a weighted sum of its past values going back to certain number of periods, together with a random disturbance in the current period. The number of past periods included in the model is denoted by p; therefore the process is denoted as AR (p). Auto regressive equation can be presented as follows: y t = Q 1 y t-1 + Q 2 y t-2 +..Q p y t-p + α +e t. (22) Where: y t = The current observation Q 1.Q p = Autoregressive coefficients p = number of previous periods included in the model α = constant term e t = an error term at time t t = the present time y t is a p th order autoregressive or AR (p) process Moving average model (MA) The process y t in this model is described by a weighted sum of current and lagged random disturbances. The time period referred back for random disturbances is denoted by q giving a process which is MA (q).

37 18 The equation of the model can be written as follows; y t = μ + e t + β 0 u t + β 1 u t-1 + β 2 u t β q u t-q.. (23) Where; μ = Constant value e = random disturbance term β 0 β q = Parameters which may be positive or negative In short, a moving average process is simply a combination of error terms Mixed autoregressive moving average model (ARMA) This model combines Autoregressive and Moving average processes altogether. In this case y t is a function of both lagged random disturbances and its past values as well as a current disturbance term. The processes AR and MA included in the model are referred as p and q respectively forming (p, q) process. The equation for the Mixed Autoregressive Moving average Model is presented as follows; y t = α +e t + Q 1 y t-1 + β 0 u t + β 1 u t-1.. (24) Causal forecasting methods This forecasting method is categorized into simple regression, multiple regression and econometric models Simple regression methods The technique tries to explain the relationship that exists between dependent and independent variables. These variables can be expressed Y = f(x), which states that

38 19 the value of Y (dependent variables) is a function of the value of X (independent variable), and is assumed to be a linear relationship. Simple regression model is represented by the function: Y = a + bx + µ. (25) Where: Y = dependent variable X = independent variable a, b = regression coefficients or parameters to be estimated µ = disturbance term Multiple regressions This is a causal forecasting method which explains the relationship between a dependent variable and two or more independent variables. Multiple Regression method is expressed in the form of equation as follows: Y = a + b 1 X 1 + b 2 X 2 +.b p X p + µ. (26) Where: a, b 1.b p = regression coefficients (parameters to be estimated) X 1..X p = independent variables p = number of variables in the equation µ = disturbance term In this study Autoregressive and smoothing models will be used for forecasting consumption of wood products in the building and construction industry. This is because the past behavior of the variable is explored to fit in the model. In this case consumption estimates of the previous years in the industry will be explored to fit in the forecasting model.

39 Econometric models Economic forecasting techniques are based on the concept that changes in economic activity can be explained by asset of mathematical relationships between economic variables. A developed theory on these relationships leads to specification of a mathematical model that expresses the nature of the relationship between the variable to be explained (dependent variable) and a set of explanatory (independent) variables. The parameters of the model will then be estimated on the basis of time series or cross sectional data. The model thus takes the form of an equation (s) that seems to be the best for describing the observed set of relationships and is in conformity with economic theory and statistical analysis. The future causes of the dependent variable will then be estimated on the basis of estimated relationships (Gregory et al., 1971).

40 21 CHAPTER THREE 3.0 METHODOLOGY 3.1 Description of the study area Location Dar es Salaam Region is located between latitudes 6 36 'and 7 South and longitudes 33 33' and 39 East. It is bordered by the Indian Ocean on the East and by the Coast Region on the other sides. Administratively, Dar es Salaam is divided into 3 municipalities namely Ilala, Kinondoni and Temeke with 73 wards altogether Area, climate and landforms Area Dar es Salaam has the total surface area of 1800 square kilometers of which 1393 square kilometers is land mass with eight offshore islands, which is about 0.19% of the entire area in Mainland Tanzania. Temeke (786.5 km 2 ) Municipality has the largest land surface area followed by Kinondoni (531 km 2 ) while Ilala (273 km 2 ) has the smallest area (Dar City Council, 2004) Climate The City experiences a modified type of equatorial climate. It is generally hot and humid throughout the year with an average temperature of 29ºC. It experiences the hottest season from October to March where temperature raises up to 35ºC. The city is moderately cool between May and August, with a temperature of about 25ºC. The city experiences two main rain seasons; a short rain season from October

41 22 to December and a long rain season between March and May. The average rainfall is 1000 mm with the lowest being 800 mm and the highest 1300 mm. Humidity is approximately 96% in the mornings and 67% in the afternoon. The climate is also influenced by the south-west monsoon winds from April to October and north-west monsoon winds between November and March. The City is divided into three ecological zones, which are the upland zone comprising of the hilly areas in the west and north, the middle plateau, and the low lands including Msimbazi valley, Jangwani, Mtoni, Africana and Ununio areas. The main natural vegetation includes coastal shrubs, miombo woodland, coastal swamps and mangrove trees (Dar City Council, 2004) Population According to 2002 national census, Dar es Salaam had a population of people. The Tanzania National Bureau of Statistics reported that the population was expected to rise to more than 3 million people by 2011 (NBS, 2006). Its population density is also reported to be 2238 per km 2 (Unhabitat, 2009) Socio- economic activities Dar es Salaam city accommodates about 40% of the total manufacturing industrial units in the country contributing about 45% of Tanzania s gross industrial manufacturing output (Unhabitat, 2009). The city is endowed with a major harbour and is considered to be an epicenter for manufacturing industry. The city attracts commercial and transportation activities from both formal and informal sectors. Increasing rates of unemployment and underemployment plays a great role in the growth of the informal sector and settlements.

42 Data collection Data for this study were collected using questionnaires, one to one interviews as well as direct observation to collect information on the consumption of sawnwood and its substitution in the building industry. Buildings from both residential and none residential areas were sampled to assess the amount of sawnwood and none wood building materials used for construction and the extent of substitution Sampling procedure The sampling was done with replacement depending on the availability of respondents and willingness to respond. From each municipality, records of registered timber traders and furniture makers were obtained. Building contractors and architects were identified through registers obtained from their respective registration boards, while house builders were identified at the building sites. The sample sizes were determined differently depending on the population of the target groups in these municipalities and the easy of accessibility. For small population comprising building contractors and public sector organizations, sample sizes of 100% were considered to increase precision. While in large population comprising traders in building materials a sample size of not less than 30% was considered as recommended by (Openshaw, 1971).In each municipality a list of Wards was prepared and 50% of wards from each municipality were sampled to ensure adequate representation of building categories. The buildings were sampled depending on their categories and the sample size ranged from 20-50% starting from lower to higher categories to increase reliability and precision.

43 Primary data collection Primary data were collected using questionnaires, data sheets, checklists and direct observation. Sawnwood used and those substituted in building activities were assessed and the amount used were estimated. Sawnwood materials used in doors and window frames were estimated in terms of volumes (m 3 ). Nonwood building materials such as aluminium were quantified and reported in square meters (m 2 ). Sawnwood volumes required for 1 m 2 of a window (0.031 m³) and 1 m 2 of a door (0.071 m³) were established from different sizes. The volumes were used as standards for other windows and doors assessed in the field. The sizes of windows and doors were measured using measuring tapes. Factors underlying sawnwood substitution were identified from each targeted group using structured and unstructured questionnaires. Residential and none residential categories were the main building strata for this study. However, the main challenge for this building s stratification was the fact that most of the buildings had multiple use ie both residential and commercial purposes. The classification of the buildings were based on their sizes and purpose to estimate the amount and type of sawnwood products and substitute materials consumed in each category. Furthermore, buildings with 4 storeys and above were under higher category while those with 1-3 storeys were under middle category. The lower category included all none storey buildings. On the other hand, none residential buildings included commercial, education, health, industries and office buildings. These subcategories were also classified into retail and storey buildings depending on their sizes for easy estimation of sawnwood

44 25 products and substitute materials consumed. Sawnwood substitution were assessed in new dwellings and in joinery (doors and windows) to see which areas are significantly substituted and the amount used Questionnaires Semi structured questionnaires with both open and closed ended questions were used to acquire both quantitative and qualitative information from each targeted groups. Questionnaires were pretested to municipal engineers of Ilala, Temeke and Kinondoni, National construction council, and Tanzania building agency leading to modification of questions number 10, 11 and 12 in questionnaire 1a to obtain reliable information. Data sheet was pilot tested to different types and size of doors and windows in all categories of buildings to prepare measurement standards for doors and windows respectively. In these questionnaires data on estimates of sawnwood products and building substitution materials, factors underlying sawnwood substitution, species, uses, availability, sources, preferences and prices of sawnwood products and substitute materials were collected Direct observation Direct observations were used during the survey and it enabled the researcher to see to what extent substitution of building materials was taking place in the buildings. This was done through visiting the construction sites to the sampled wards and streets. Information on consumption of sawnwood products, species used, and substitute building materials were captured and recorded in special data sheet.

45 Checklists and Key Informants The key informants in this study were construction engineers, building contractors, house builders, forest officers, town planners, quantity surveyors, site inspectors and land officers. Personal interviews were conducted to acquire more information on quantities, type of permits, size of plots, and building permits issued for previous years. Key informants were selected in a way that they can provide relevant information on rate of consumption and substitution, species used in the building construction and the factors underlying wood substitution in the building and construction activities. Tanzania building agency (TBA) offices which is responsible for building government houses were visited to obtain reliable information on use of sawnwood and other materials used in the buildings Secondary data collection Secondary data on consumption of wood products, substitute building materials and factors underlying wood substitutions and consumption in the building sector were collected from the three municipalities of Ilala, Temeke and Kinondoni. Other secondary information was accessed from different documents available in the municipalities, official publications, reports and journals. 3.3 Data analysis Both quantitative and qualitative data were, coded, checked and analysed using Statistical Package for Social Sciences (SPSS) and MS-excel computer programme. Verbal responses from key informants were analysed using content analysis to get the meaningful information. The output were summarized in a descriptive statistics

46 27 such as frequency, mean and sums especially on the uses, availability, sources, preferences, of wood products and substitute building materials. The mean and sums of quantities of sawnwood products and substituted building materials consumed and prices were also obtained Wood products consumption forecast Forecasting consumption of sawnwood and the substituted materials in the building industry were done. Forecasting for years 2016, 2021 and 2026 were performed using the model that fitted the collected data. Before using any model the forecasting techniques were evaluated to get the appropriate forecasting model Smoothing (average) technique Basing on the previous consumption of sawnwood products in the building industry single moving average and single exponential smoothing models were used. Single moving average technique was chosen to compute forecast for future periods because it fits with the limited amount of data obtained from the study. Data requirement for application of single moving average are minimal. The accuracy of this method in computing forecast for future periods is low compared to single exponential smoothing which gives more weight to the recent observations. Recent observations contain more information pertaining the future than older ones and therefore single exponential smoothing takes care of the weaknesses that are shown by single moving average technique. This technique can be developed starting from the single moving average equation as illustrated by Wheelwright and Makridakis (1985).